We’re currently using the Google Vertex Chat Model node in n8n and are trying to use Google’s newly released model gemini-3.1-flash-lite (GA on Vertex AI as of May 7, 2026).
When entering gemini-3.1-flash-lite as the model name, we get the error: “No model found called ‘gemini-3.1-flash-lite’”
Our setup:
Credential: Google Service Account (connection tested successfully )
Region: europe-west4
Model name: gemini-3.1-flash-lite
Note: gemini-2.5-flash works fine with the exact same credential and region, so the issue is specifically with the 3.1 Flash-Lite model name not being recognized by the node.
Could you confirm:
Is gemini-3.1-flash-lite currently supported in the Google Vertex Chat Model node?
If not, is there a planned timeline for adding support?
Yes, I think there is a problem related to the dropdown. Have you tried pasting this exact name in the model name {{ 'gemini-3.1-flash-lite' }} ? As if the endpoint had it, then it should work.
Welcome to the n8n community @lenolo1815
I understand that the post was about the limitation for the latest gemini and that you were able to use another version of it. I’d like to add that if possible, you can call the 3.1 through an http request, while this update isn’t available yet.
@lenolo1815
I didn’t find anything on the subject, but since it’s just the beginning of the month and knowing there are several processes that need to be completed for this to be enabled, I recommend waiting.
Is this affecting the overall process too much? Why the dependency on this specific model?
Because it’s the most reliable model for AI OCR and we need to host in Europe for GDPR compliance. We can wait, but isn’t there roughly a timeframe for how long something like that takes?
Hello, there’s the same issue with gemini-3.5-flash. I think the main problem is that the eu-west-…, eu-east-… etc. localizations are no longer available for gemini 3 models, they’ve been replaced with “eu” or “na” localizations.
Is updating the Google Vertex Chat Model node planned in the roadmap to account for this change?
I can’t speak to development, but the solution from @nguyenthieutoan absolutely did the trick for me. If you are coming to this post, it appears that the issue for being able to access models - via chat - is the region on your service account credential.
I will look (or post) for other discussions around missing features relative to the Gemini nodes and what is possible with models run through vertex (now AI Studio?)
Okay, just pointing out a potential issue for everyone who needs to be GDPR compliant:
google will discontinue 2.5 models around october 2026 (see here)
the new 3.5 models ARE NOT supported in individual eu regions (e.g. “europe-west4”) (see here)
google has introduced a new eu-region endpoint called “eu” (see here)
the new 3.5 models ARE supported in this eu-region
BUT the this new eu-region endoint needs a different SDK endpoint url format (see here), so its not compatible with the current n8n nodes
So there IS a bit of time-pressure on this, since otherwise all n8n Setups that need to be GDPR compliant will fail in october without any way to fix this.
As discovered by developers tracking this exact bug in the n8n community, the fastest fix is to switch the region on your Google Vertex node/credential setup.
Change your region from europe-west4 to us-central1.
The gemini-3.1 flash lite production API endpoints are most stable and universally mapped under the primary US regions. Your service account credentials will still authenticate perfectly fine regardless of the region shift.
)
